Sensory Perception Systems and Machine Learning Methods for Pesticide Detection in Fruits
Round 1
Reviewer 1 Report
Comments and Suggestions for AuthorsThis paper describes a pilot study that was carried out to evaluate the ability to detect pesticides in fruits through an Electronic Nose and Electronic Tongue
Some minor changes need to be fixed before publication
- A graphical abstract associated with the paper will improve the impact on the readers
- Table 1- difficulties to see the structures – colours, height of letters
- Pag 7 – line 218 - 219 – more details about chamber of 16 Taguchi-type gas sensors - 218 manufactured by Figaro sensor company are necessary to be included in the article. The limit of detections needs to be added in Table 2
- Figure 3- 20 – unclear – numbers, axes.. The explanations associated are not comments even after the figure and are also not in the discussion sub-chapter. More explanations are needed
Comments on the Quality of English Languageminor changes
Author Response
Reviewer’s comments
We appreciated the editors' and reviewers' remarks, which helped us improve our manuscript. Based on their suggestions and comments, we have made several changes, using different colors to highlight the corrections.
- We have significantly improved the abstract to offer a more concise and informative overview of the study, ensuring it captures the essence and significance of the research.
- The descriptions and explanations throughout the document have been enhanced for clarity and depth, providing readers with a better understanding of the study’s context and findings.
- Figures and tables have been revised with clearer labels and improved data presentation to support the text more effectively.
- We have explicitly highlighted the study’s aims and advantages.
Reviewer 1:
This paper describes a pilot study that was carried out to evaluate the ability to detect pesticides in fruits through an Electronic Nose and Electronic Tongue
Some minor changes need to be fixed before publication
- A graphical abstract associated with the paper will improve the impact on the readers.
Response: Thank you for your suggestion. Graphical abstract has been included.
- Table 1- difficulties to see the structures – colours, height of letters.
Response: Thank you for your suggestion. We have improved the Table 1.
- Pag 7 – line 218 - 219 – more details about chamber of 16 Taguchi-type gas sensors - 218 manufactured by Figaro sensor company are necessary to be included in the article. The limit of detections needs to be added in Table 2.
Response: Thank you for your suggestion. We have described the sensor chamber and detection limit overall (Table 2).
A multisensory system was developed to detect pesticide residues in fruits. This system consists of a concentration chamber and a measurement chamber. The concentration chamber is constructed from durable methacrylate and has dimensions of 12 cm in length, 12 cm in width, and 12 cm in height, resulting in a total internal volume of 1728 cm³. The measurement chamber, also made from methacrylate, contains 16 metal oxide gas sensors, specifically Taguchi sensors from the Figaro Company, located in Arlington Heights, IL, USA. This chamber measures 8.7 cm in length, 8.7 cm in width, and 5.3 cm in height, with an internal volume of 401.157 cm³. The sensors are capable of detecting a wide range of compounds, including ammonia, amines, monochlorodifluoromethane, hydrogen, methane, volatile gases, water vapor in food, hydrogen sulfide, hydrocarbons, carbon monoxide, iso-butane, ethanol, alcohol, organic solvents, and tetrafluoromethane (as detailed in Table 2). Figure 2 illustrates the measurement setup for the E-nose system.
- Figure 3- 20 – unclear – numbers, axes.. The explanations associated are not comments even after the figure and are also not in the discussion sub-chapter. More explanations are needed.
Response: Thank you for your suggestion. We have improved all the figures and included an explanation for each one.
Author Response File: Author Response.pdf
Reviewer 2 Report
Comments and Suggestions for AuthorsReview Comments
The manuscript investigates the use of electronic nose (E-nose) and electronic tongue (E-tongue) technologies for detecting pesticide residues in fruits. It explores the potential of these artificial sensory systems to differentiate between organic fruits and those treated with pesticides at various concentrations. The study employs multivariate analysis techniques such as Principal Component Analysis (PCA) and Linear Discriminant Analysis (LDA), alongside machine learning algorithms like Decision Trees, Naïve Bayes, Support Vector Machines (SVM), and K-Nearest Neighbors (KNN) to process and analyze the sensor data. However, several critical issues need to be addressed to strengthen the research and its applicability. Consequently, I suggest giving this manuscript a major revision before considering it for publication.
The detailed comments are enumerated as follows:
1. Figure 2 in the manuscript, which illustrates the operation diagram of the E-nose during the cleaning and measurement phases, suffers from issues with image and font compression that could hinder the clarity and professionalism expected in the journal’s publications. Besides, all the figures in the article are unreadably small. The text labels and diagrammatic details are not clearly visible, which might lead to ambiguities in understanding the operation mechanism described.
2. Table4 provided in the manuscript detailing the accuracy of various classification methods using both electronic nose and electronic tongue for different fruits raises significant issues regarding the clarity and completeness of the reported performance metrics. The manuscript does not specify whether the reported accuracies are from the training, validation, or testing phases. This distinction is crucial because accuracy during training can be misleading due to potential overfitting, whereas validation and testing accuracies provide a more realistic measure of how well the models generalize to new data. It is unclear whether the accuracies reported are maximum values observed during the training process or averages over multiple runs or folds. This distinction is important to evaluate the consistency and reliability of the models.
3. PCA is fundamentally a dimensionality reduction technique and not a predictive model. It is typically used to simplify data, enhance visualization, and identify patterns. However, it does not provide mechanisms for classification or prediction on its own. Relying solely on PCA may not fully capture the potential of machine learning techniques to classify or predict outcomes based on sensor data.
4. Visualization tools such as confusion matrices, learning curves, violin plots, etc. can be used to clearly show the model performance.
5. It seems the networks achieves a best result of only a little above 90% accuracy for the fairly simple task of identifying classification of fruits with pesticides- there have been much better results resented before, so the accuracies presented in this paper are not competitive.
6. There is a lack of detailed explanation about the implementation specifics, such as parameter settings, model tuning, and the handling of potential overfitting, which are critical for reproducibility and understanding the robustness of the findings.
7. While accuracy is frequently reported, other critical metrics such as precision, recall, F1-score, and AUC that provide more insight into model performance, especially in imbalanced datasets, are not discussed. Reporting these metrics would provide a more comprehensive evaluation of the models.
8. Currently, deep learning models and multi-task techniques are widely used and may be considered for comparison with the traditional machine learning methods used in the manuscript. There are many studies about the deep learning methods of the electronic nose for VOCs detection. I suggest the authors have some review or discussion of these articles if appropriate.
e.g., [1] W. Ni, T. Wang, Y. Wu, X. Liu, Z. Li, R. Yang, K. Zhang, J. Yang, M. Zeng, N. Hu, B. Li, Z. Yang, Multi-task deep learning model for quantitative volatile organic compounds analysis by feature fusion of electronic nose sensing, Sens. Actuators B: Chem. 417 (2024) 136206. https://doi.org/10.1016/j.snb.2024.136206.
Comments on the Quality of English LanguageEnglish language is fine.
Author Response
Reviewer’s comments
We appreciated the editors' and reviewers' remarks, which helped us improve our manuscript. Based on their suggestions and comments, we have made several changes, using different colors to highlight the corrections.
- We have significantly improved the abstract to offer a more concise and informative overview of the study, ensuring it captures the essence and significance of the research.
- The descriptions and explanations throughout the document have been enhanced for clarity and depth, providing readers with a better understanding of the study’s context and findings.
- Figures and tables have been revised with clearer labels and improved data presentation to support the text more effectively.
- We have explicitly highlighted the study’s aims and advantages.
Reviewer 2:
We appreciated the editors' and reviewers' remarks, which helped us improve our manuscript. Based on their suggestions and comments, we have made several changes.
The manuscript investigates the use of electronic nose (E-nose) and electronic tongue (E-tongue) technologies for detecting pesticide residues in fruits. It explores the potential of these artificial sensory systems to differentiate between organic fruits and those treated with pesticides at various concentrations. The study employs multivariate analysis techniques such as Principal Component Analysis (PCA) and Linear Discriminant Analysis (LDA), alongside machine learning algorithms like Decision Trees, Naïve Bayes, Support Vector Machines (SVM), and K-Nearest Neighbors (KNN) to process and analyze the sensor data. However, several critical issues need to be addressed to strengthen the research and its applicability. Consequently, I suggest giving this manuscript a major revision before considering it for publication.
The detailed comments are enumerated as follows:
- Figure 2 in the manuscript, which illustrates the operation diagram of the E-nose during the cleaning and measurement phases, suffers from issues with image and font compression that could hinder the clarity and professionalism expected in the journal’s publications. Besides,all the figures in the article are unreadably small. The text labels and diagrammatic details are not clearly visible, which might lead to ambiguities in understanding the operation mechanism
Response: Thank you for your comment. Figure 2 was improved regarding the image and font.
- Table 4 provided in the manuscript detailing the accuracy of various classification methods using both electronic nose and electronic tongue for different fruits raises significant issues regarding the clarity and completeness of the reported performance metrics.The manuscript does not specify whether the reported accuracies are from the training, validation, or testing phases. This distinction is crucial because accuracy during training can be misleading due to potential overfitting, whereas validation and testing accuracies provide a more realistic measure of how well the models generalize to new data. It is unclear whether the accuracies reported are maximum values observed during the training process or averages over multiple runs or folds. This distinction is important to evaluate the consistency and reliability of the models.
Response: Thank you for your suggestion. We have explained Table 4 regarding the reported accuracies.
It is important to clarify that the accuracy achieved using ML methods based on ET and EN systems was initially derived from the training and validation phases, where the largest possible dataset was utilized. This approach was taken to develop models that avoid overfitting, ensuring that they generalize well. These models were subsequently applied in the testing phase using a new set of measurements to evaluate their performance on unseen data (see Section 3.5). The careful selection and use of a comprehensive dataset during the initial phases were crucial in building robust models capable of reliable performance when faced with new samples.
- PCA is fundamentally a dimensionality reduction technique and not a predictive model. It is typically used to simplify data, enhance visualization, and identify patterns. However, it does not provide mechanisms for classification or prediction on its own. Relying solely on PCA may not fully capture the potential of machine learning techniques to classify or predict outcomes based on sensor data.
Thank you for your suggestion. We have made the following explanation.
It is important to recognize both the specific role that PCA plays and its broader impact on the predictive modeling process, especially in the context of sensor data analysis. In this study, we have used PCA analysis to transform the original dataset into a set of uncorrelated principal components that retain the most significant information by reducing the dimensionality, where PCA not only simplifies the dataset but also enhances the performance of predictive models by mitigating issues like multicollinearity, which can hinder the model's ability to learn effectively from our dataset. In our analysis, we employed PCA primarily to visualize the behavior and discrimination of pesticide categories in fruits within a two-dimensional space. It is important to note that PCA was not used to classify the measurements. However, its application has also been demonstrated in optimizing machine learning methods, as it enabled extracting more meaningful and informative features from the data, thereby enhancing the performance of the predictive models.
- Visualization tools such as confusion matrices, learning curves, violin plots, etc. can be used to clearly show the model performance.
Thank you for your suggestion.
We have included Tables 5 and 6, which were developed using confusion matrices to evaluate the performance of the models. These tables detail how well our models classify different categories for fruit pesticide identification using E-Nose and E-Tongue systems. The confusion matrices enable us to compute key performance metrics, such as Accuracy, precision, Sensitivity, Specificity, F1-score and NPV, offering a comprehensive assessment of model effectiveness.
We opted not to include the confusion matrices in the paper to keep the overall length manageable. Including these additional details would have significantly increased the paper's size, potentially detracting from the clarity and focus of our key findings.
Table 5 presents the metrics used to evaluate the performance of an E-nose in discriminating between organic fruits and fruits contaminated with various pesticides, based on the selection of the best classification model obtained in Table 4 and the confusion matrix. The metrics presented include precision, sensitivity, specificity, accuracy, F1-score, and negative predictive value (NPV). The Linear Discrimination model, in most cases, achieved classification values above 90%, with precision rates reaching up to 100% in identifying pesticides in some fruits. The E-Tongue appears to be more effective than the E-nose for classifying pesticide-treated fruits. This is evidenced by its greater accuracy and consistency across all classification methods and fruit types. The E-nose, although a useful system for agro-food applications, shows a more variable performance and, in this study, was less effective on fruits such as strawberries and cape gooseberries, indicating that the volatile compounds detected by the electronic.
Likewise, Table 6 illustrates the metrics derived from the confusion matrix of the SVM, Linear Discriminant, and Naïve Bayes models, which show precision values reaching up to 100% in identifying nearly all pesticides in fruits. This result confirms that the E-Tongue outperformed the E-nose, demonstrating superior accuracy and effectiveness in pesticide detection across various fruit samples.
- It seems the networks achieves a best result of only a little above 90% accuracy for the fairly simple task of identifying classification of fruits with pesticides- there have been much better results resented before, so the accuracies presented in this paper are not competitive.
Thank you for your suggestion. We have provided an extensive explanation regarding this comment.
In the study conducted by Rivera Nategh et al. [72], an E-nose composed of eight chemoresistive sensors was used to detect organophosphate pesticide residues in peaches (Prunus persica) and distinguish between three concentrations (1, 2, and 3 ppm) as well as pesticide-free fruit. The sensor responses were recorded using PCA, achieving a variance of 99.8% with two principal components. Another study by Nategh et al. [71] demonstrated the capability of an E-nose composed of 10 MQ and TGS reference sensors to detect Diazinon pesticide residues in cherries and differentiate between four stages of fruit maturity. The authors employed three data analysis methods: PCA, LDA, and ANN. PCA score plots of PC1–PC2 captured 90-96% of the variance in the data for toxic and non-toxic sweet cherries, while ANN and LDA achieved 100% classification accuracy. In another study, Tang et al. [70] evaluated the performance of a commercial PEN3 E-nose equipped with 10 metal oxide semiconductor (MOS) sensors for detecting pesticides in apple samples. The fruit samples were treated with two pesticides containing cypermethrin and chlorpyrifos at four different concentrations and one mixture. PCA and LDA methods successfully differentiated between control and contaminated samples at varying concentrations. Although both methods were effective, PCA exhibited the best discrimination ability. The SVM method demonstrated high accuracy, ranging from 94% to 97% for training datasets and from 90% to 93% for testing datasets, concluding that the commercial E-nose is a useful tool for pesticide residue detection. In a more recent study, an E-nose composed of 11 gas sensors was fabricated, allowing for the differentiation between pesticide-contaminated cherries (Korban) and non-contaminated cherries. Four classification algorithms (Random Forest Classifier, Extra Trees Classifier, Decision Tree Classifier, and k-nearest Neighbor (K-NN)) were employed, with the Extra Trees Classifier achieving the most satisfactory results, with a classification accuracy of 94.30%, a sensitivity of 93.00%, and a specificity of 95.60%[73]. Regarding E-Tongues, there is limited literature focused on pesticide detection in fruits. However, some biosensors made from various materials have been used to detect pesticides (malathion and cadusafos) in laboratory-prepared matrices (water, food). In terms of classification, sensitivity, and specificity, studies on pesticide detection in fruits using E-noses show promising results. Additionally, the ability of these devices to differentiate between various types of pesticides and concentrations, along with their high sensitivity and specificity, makes them promising tools for enhancing safety and quality control in the food industry[74-76]. However, to date, no studies have been reported in the literature that implement both an E-nose and tongue simultaneously to detect pesticides in different fruits and concentrations. Only one study conducted by the same authors [77] managed to discriminate and classify newly cultivated and organic fruits using both technologies. Both devices demonstrated the ability to discriminate and classify the studied samples, although the E-Tongue exhibited greater data dispersion and some overlap between the analyzed data classes. Therefore, the distinction between the study proposed in this research and the others mentioned above lies in using both technologies to detect pesticides in various fruits characteristic of the region compared to organic fruit that was free of contamination. Additionally, different performance metrics of the E-nose and tongue were evaluated using the best classification model reported in Table 4, followed by the evaluation of the metrics for the best models. Overall, both devices achieved up to 100% classification accuracy in detecting pesticides in some fruits, but the E-Tongue showed better precision in most cases, particularly for fruits like plums, apples, and cape gooseberries treated with various pesticides, indicating a robust capability to correctly identify both organic and pesticide-treated fruits. It is also evident that in other metrics (sensitivity, accuracy, and F1-score), the E-Tongue demonstrated better efficiency than the E-nose, as the latter showed more variability in the data, especially for strawberries and cape gooseberries. This could suggest that the E-nose is more sensitive to variations in the fruit matrix or pesticide characteristics. Additionally, different pesticide concentration ranges were evaluated to assess the ability of these devices to detect and classify samples within categories, generally achieving accuracy greater than 80% and reaching 100% in detecting Preza and Bricol for cape gooseberries, depending on the classifier implemented. Similarly, the E-Tongue showed better performance than the E-nose in discriminating pesticide concentrations in fruits. This is attributed to its higher precision, sensitivity, accuracy, and specificity, as well as its more effective capability to distinguish between different pesticide concentrations in various fruits. Finally, test samples (whose condition was previously unknown) were used with the model created in the first stage of this study to determine the type of pesticide contained in the test samples, where successful classification results were achieved.
Although good results have been obtained, it is necessary to validate the operation of the sensory equipment based on the previously created model and trained with the learning methods, where it is also convenient to use commercial analytical equipment such as GC-MS, since it allows the search for the active ingredients of pesticides, and in this way verify the results obtained previously. Therefore, the detection of the active ingredients of pesticides employing this methodology would help us to confirm the information provided by farmers and by the responses of sensory perception systems.
Based on the results obtained in this study, we achieved high accuracies, with some classifications reaching 100%. These outcomes demonstrate the effectiveness of our approach and make our study competitive when compared to previous research in the field of pesticide classification
- There is a lack of detailed explanation about the implementation specifics, such as parameter settings, model tuning, and the handling of potential overfitting, which are critical for reproducibility and understanding the robustness of the findings.
Thank you for your remark. We have provided an overall explanation of this remark.
It was crucial to undertake a comprehensive process involving several key steps to effectively utilize machine learning algorithms through this tool for detecting fruit pesticides. First, a feature extraction technique based on two parameters was obtained for each signal: the maximum (max) and minimum (min) values of the signal (i.e., voltage for gas sensor signals and current for electrochemical responses), which represent the highest and lowest points that the signal reaches over a given period. This value provided a quantitative measure of the signal’s variability, which was subsequently used for signal analysis by applying PCA and LDA methods and classification techniques, as described above. On the other hand, data normalization was applied to the data set using the “mean-centering” and “Auto-scaled” algorithms to ensure that each feature contributes equally to the model. Then, these parameters were calculated by subtracting the mean or dividing by the standard deviation for each feature. Afterward, it was necessary to optimize the parameters of ML methods (Decision Tree classifiers, Linear Discrimination, Naïve Bayes, quadratic SVM, cubic SVM, and Fine K-NN) and select appropriate hyperparameters (i.e., kernel) for each method, such as cubic SVM and the Fine K-NN, learning rates, and regularization parameters. Consequently, the models were refined through a meticulous process of fine-tuning. This involved iteratively adjusting the models based on feedback from initial training and validation runs to enhance their predictive accuracy and generalization capabilities.
The training data was randomly divided into multiple subsets to validate the model's performance across different segments, and regularization helped control model complexity. For example, whether k=5, the dataset is split into 5 equal parts. Therefore, k-fold cross-validation divides the dataset into k subsets, and the model is trained and tested k times, each time using a different fold as the test set and the remaining folds as the training set. Afterward, test data was used with the trained model to assess the final model's performance and generalization ability. This data was not used in any way during the training or validation phases.
- It was crucial to undertake a comprehensive process involving several key steps to effectively utilize machine learning algorithms through this tool for detecting fruit pesticides.
Thank you for your remark. We have explained this remark overall (Please, reference the previous text), and:
The methods outlined in the previous sections were implemented using the "Classification Learner" interface in MATLAB Version 2020b. This tool significantly streamlined the development process for classification models, providing an intuitive and efficient platform for selecting features, training models, and evaluating performance metrics.
- While accuracy is frequently reported, other critical metrics such as precision, recall, F1-score, and AUC that provide more insight into model performance, especially in imbalanced datasets, are not discussed. Reporting these metrics would provide a more comprehensive evaluation of the models.
Thank you for your comment. We have provided information related to metrics using ML methods.
Table 5 presents the metrics used to evaluate the performance of an E-nose in discriminating between organic fruits and fruits contaminated with various pesticides, based on the selection of the best classification model obtained in Table 4 and the confusion matrix. The metrics presented include precision, sensitivity, specificity, accuracy, F1-score, and negative predictive value (NPV). The Linear Discrimination model, in most cases, achieved classification values above 90%, with precision rates reaching up to 100% in identifying pesticides in some fruits. The E-Tongue appears to be more effective than the E-nose for classifying pesticide-treated fruits. This is evidenced by its greater accuracy and consistency across all classification methods and fruit types. The E-nose, although a useful system for agro-food applications, shows a more variable performance and, in this study, was less effective on fruits such as strawberries and cape gooseberries, indicating that the volatile compounds detected by the electronic.
Likewise, Table 6 illustrates the metrics derived from the confusion matrix of the SVM, Linear Discriminant, and Naïve Bayes models, which show precision values reaching up to 100% in identifying nearly all pesticides in fruits. This result confirms that the E-Tongue outperformed the E-nose, demonstrating superior accuracy and effectiveness in pesticide detection across various fruit samples.
- Currently, deep learning models and multi-task techniques are widely used and may be considered for comparison with the traditional machine learning methods used in the manuscript. There are many studies about the deep learning methods of the electronic nose for VOCs detection. I suggest the authors have some review or discussion of these articles if appropriate.
e.g., [1] W. Ni, T. Wang, Y. Wu, X. Liu, Z. Li, R. Yang, K. Zhang, J. Yang, M. Zeng, N. Hu, B. Li, Z. Yang, Multi-task deep learning model for quantitative volatile organic compounds analysis by feature fusion of electronic nose sensing, Sens. Actuators B: Chem. 417 (2024) 136206. https://doi.org/10.1016/j.snb.2024.136206.
Thank you for your remark and advice. We have not used this technique because these kinds of models typically require large amounts of data to generalize well. In our case, we had small datasets; therefore, deep learning models can be disposed to overfitting, where the model learns noise and details specific to the training data rather than general patterns.
Author Response File: Author Response.pdf
Reviewer 3 Report
Comments and Suggestions for AuthorsThis manuscript utilizes electronic nose and electronic tongue combined with various chemometric methods to identify organic fruits and fruits with pesticide residues, further distinguishing fruits containing different concentrations of insecticides and fungicides. It demonstrates a certain level of innovation and application prospects, but the following problems need to be improved.
(1) The logical structure of the article still requires further organization and refinement. For example, the abstract does not mention the advantages of electronic nose and electronic tongue. The data processing section does not display on the software used for handling data, and the method of dividing the training set and validation set. The results section needs to integrate and streamline different methods and provide corresponding explanations.
(2) Line 20: Please add the advantages of the electronic tongue and electronic nose techniques compared to traditional techniques.
(3) Line 21: Please add the names of the fruits and the types of pesticides to be identified.
(4) Suggest integrating and streamlining paragraphs 2, 3, 4, and 5 of the introduction section to emphasize the research aim.
(5) Line 163-164, Please explain the aim and significance of the study.
(6) Line 171, It is recommended to integrate this overall scheme into the research purpose of the introduction section.
(7) Line 175, Please provide more details on why chose the four fruits from these regions.
(8) Line 281, Please clarify the specific data processing software and the method of dividing the training set and the test set in supervised algorithm.
(9) Line 335 and 350, Please explain why? Add the LDA identification accuracies in correspondent section.
(10) Please refer to the annotations in the manuscript for other issues.
Comments for author File: Comments.pdf
Language can be understood, but it can still be improved.
Author Response
Reviewer’s comments
We appreciated the editors' and reviewers' remarks, which helped us improve our manuscript. Based on their suggestions and comments, we have made several changes, using different colors to highlight the corrections.
- We have significantly improved the abstract to offer a more concise and informative overview of the study, ensuring it captures the essence and significance of the research.
- The descriptions and explanations throughout the document have been enhanced for clarity and depth, providing readers with a better understanding of the study’s context and findings.
- Figures and tables have been revised with clearer labels and improved data presentation to support the text more effectively.
- We have explicitly highlighted the study’s aims and advantages.
Reviewer 3:
We appreciated the editors' and reviewers' remarks, which helped us improve our manuscript. Based on their suggestions and comments, we have made several changes.
This manuscript utilizes electronic nose and electronic tongue combined with various chemometric methods to identify organic fruits and fruits with pesticide residues, further distinguishing fruits containing different concentrations of insecticides and fungicides. It demonstrates a certain level of innovation and application prospects, but the following problems need to be improved.
- The logical structure of the article still requires further organization and refinement. For example, the abstract does not mention the advantages of electronic nose and electronic tongue. The data processing section does not display on the software used for handling data, and the method of dividing the training set and validation set. The results section needs to integrate and streamline different methods and provide corresponding explanations.
Thank you for your remark. We have organized the structure of our manuscript.
Abstract section: In this study, an electronic tongue (ET) and electronic nose (EN) systems were implemented in detecting pesticide residues such as Preza, Daconil, Curzate, Bricol, Accros, Amistar, and Funlate in cape gooseberry, apples, plums, and strawberries fruits, which offers several advantages over traditional analytical methods, in terms of speed, cost, ease of use, portability, non-destructive analysis, real-time monitoring, versatility, and environmental impact. These benefits make ET and EN particularly well-suited for routine screening, on-site analysis, and applications where rapid, cost-effective detection of pesticide residues is critical.
2.5 Data processing: The methods outlined in the previous sections were implemented using the "Classification Learner" interface in MATLAB Version 2020b. This tool significantly streamlined the development process for classification models, providing an intuitive and efficient platform for selecting features, training models, and evaluating performance metrics.
- Line 20: Please add the advantages of the electronic tongue and electronic nose techniques compared to traditional techniques.
Thank you for your comment. We have included the advantages of both systems.
In this study, we have implemented an electronic tongue (E-Tongue) and electronic nose (E-Nose) techniques in detecting pesticide residues such as Preza, Daconil, Curzate, Bricol, Accros, Amistar, and Funlate in cape gooseberry, apples, plums, and strawberries fruits, which offers several advantages over traditional analytical methods, in terms of speed, cost, ease of use, portability, non-destructive analysis, real-time monitoring, versatility, and environmental impact.
- Line 21: Please add the names of the fruits and the types of pesticides to be identified.
Thank you for your comment. We have added the names of the fruits and Types of pesticides (see abstract).
We have implemented an electronic tongue (ET) and electronic nose (EN) techniques in detecting pesticide residues such as Preza, Daconil, Curzate, Bricol, Accros, Amistar, and Funlate in cape gooseberry, apples, plums, and strawberries fruits.
- Suggest integrating and streamlining paragraphs 2, 3, 4, and 5 of the introduction section to emphasize the research aim.
Thank you for your remark. We have improved the paragraphs of 2,3,4 and 5.
- Line 163-164, Please explain the aim and significance of the study.
Thank you for your comments We have explained the aim of our study.
This study aims to validate the use of E-Tongue and E-Nose technologies as innovative tools for detecting and classifying pesticide residues in various fruits. The fruits selected are among the most widely cultivated in the Santander region, making them ideal candidates for evaluating the effectiveness and practicality of these devices as alternatives to traditional pesticide detection methods, such as GC-MS. The evaluation focused on key advantages of ET and EN, including non-destructive analysis, portability, speed, and cost-effectiveness. The study began with training and validation phases, followed by testing with new data to assess the devices' capabilities. The analysis revealed high accuracy in classifying both the presence of pesticide residues and their concentrations using multivariate analysis and advanced Machine Learning (ML) techniques. This is particularly significant as it meets the growing demand for more accessible, efficient, and environmentally friendly methods to monitor pesticide residues in agricultural products.
- Line 171, It is recommended to integrate this overall scheme into the research purpose of the introduction section.
Thank you for your remark. We have moved this scheme to the introduction section.
- Line 175, Please provide more details on why chose the four fruits from these regions.
Thank you for your remark. We have provided more details on selecting the fruits from the Colombia regions.
The fruits selected are among the most widely cultivated in the Santander region, making them ideal candidates for evaluating the effectiveness and practicality of these devices as alternatives to traditional pesticide detection methods, such as GC-MS.
- Line 281, Please clarify the specific data processing software and the method of dividing the training set and the test set in supervised algorithm.
Thank you for your remark. We describe the data processing software and the method of dividing the training set.
The methods outlined in the previous sections were implemented using the "Classification Learner" interface in MATLAB Version 2020b. This tool significantly streamlined the development process for classification models, providing an intuitive and efficient platform for selecting features, training models, and evaluating performance metrics. It also included techniques such as cross-validation (k-fold=5), where the training data were randomly divided into k multiple subsets to validate the model's performance across different segments, and regularization, which helped control model complexity. For example, whether k=5, the dataset is split into 5 equal parts. k-fold cross-validation divides the dataset into k subsets, and the model is trained and tested k times, each time using a different fold as the test set and the remaining folds as the training set. Afterward, test data was used with the trained validation model to assess the final model's performance and generalization ability. This data was not used in any way during the training or validation phases.
- Line 335 and 350, Please explain why? Add the LDA identification accuracies in correspondent section.
Thank you for your remark. We added the LDA identification accuracies.
The LDA method was used together in this study since PCA analysis may be used to reduce the dimensionality of the data, followed by LDA to further reduce the dimensions and optimize the separation between classes for classification tasks. The classification method resulted in a level of 90 % accuracy.
- Please refer to the annotations in the manuscript for other issues.
Thank you for your remark and contribution. We have checked the relevant annotations in the manuscript.
Author Response File: Author Response.pdf
Round 2
Reviewer 2 Report
Comments and Suggestions for AuthorsThe authors have revised the manuscript and appropriately addressed the questions in the comments. I recommend accepting the current version of the manuscript.
Reviewer 3 Report
Comments and Suggestions for AuthorsThis manuscript has been carefully modified by the authors, and the research purpose, significance and content can be clearly stated. Some small details can be modified to better improve the quality of this text.
(1)
Comments on the Quality of English Language
The English language is fine, but the readability can be better if there are proper breaks.